Listen or watch on Spotify, Apple Podcasts and Youtube.
Cal Thompson is a designer and product advisor. She was VP of Product Design and Research at Headspace, spent years in service design at Fjord, and now works as a fractional chief design officer and advisor to startups. She also wrote and teaches the Product Design Strategy course at Future London Academy. Outside of product work, she is a painter and sculptor.
One question ran through our whole conversation: when anyone can build anything, what should we actually build? In this conversation, Cal explains why engagement is an uncreative metric, why everyone on a product team now needs to learn the design process, how AI fills products with “user need hallucinations,” what human metrics can look like in practice, and why painting keeps her clear-headed as a designer.
Let’s dive in.
Losing the Thread
Cal starts with a coffee cup. You pick it up, you drink, you put it down, and you’re done. Nobody designed it to make you keep holding it.
Digital products have drifted away from that simple idea. Somewhere along the way, success came to mean that people use a product as much as possible. Cal sees that as borrowed thinking. Notification dots, likes, and the pull to come back and check were built into social media so well that many people got hooked. Young people, who are already in a vulnerable stage of life, feel it most.
I think we’ve kind of lost the thread on like what a product is meant to be for.
Copying those mechanics into every other kind of product isn’t responsible design. It isn’t even original.
So it’s almost like uncreative product metrics in a way.
Everyone Is Building Now
For a long time, designers carried much of the blame for bad product decisions. With AI-enabled building, that circle has grown. Product managers, engineers, and founders can all ship features now. Cal sees the result in the US already: apps filling up with extras nobody asked for, like a prompt offering to summarize text you never wanted summarized.
Her answer isn’t to keep others away from design. It’s the opposite. If everyone is going to play in the sandbox, everyone needs to learn how design actually works, from research to synthesis, ideation, testing, and iteration.
And so the remedy is everyone needs to get educated in design.
There’s an echo of history here. UX as a discipline grew out of a similar chaos, when Don Norman began working at Apple and developers were building everything without much thought for the human side of the interface. The tools are different now, but the problem looks familiar.
The Hammer and the Bell
Cal often shows a slide of a hammer. Everyone knows how to use one. The grip tells you where to hold it. The weight sits at the end, so it does its job. It is a tool with one clear purpose.
Now imagine that hammer with a little bell on top, a ribbon hanging off one side, and a sheet of screen dangling from the handle. That, she says, is what many products are becoming as AI-generated features pile up. When I called it kitsch, Cal laughed and agreed. Her own name for it is “user need hallucinations”: features that sound plausible but aren’t grounded in any real observed need.
If you can make a tool that is as clear in its purpose as any of these household tools that we have, then you’ve succeeded.
She points to an everyday example from her time in Lithuania. The Bolt app feels clearer to her than Uber does back home, simply because it sticks to calling a car. Uber in the US has added so much on top that the core purpose gets lost.
That’s all people need you for. We don’t need everything else.
Going Into People’s Homes
Designers, especially those coming from visual or brand design, often start from a brief. Cal’s first piece of advice is to learn observational field research before anything else.
She remembers a project for Kaiser Permanente in San Francisco. The team went into homes across the Bay Area and talked with people about their health. They did kitchen tours and fridge tours, and looked at the gadgets people used. Many people said their dog kept them healthy, because of all the walks. That kind of insight doesn’t come from a desk.
She even brought the visual designers along, people usually pulled in near the end of a project. Seeing the problems firsthand gave everyone a reason to care. You remember the woman who struggled with something, and you want to solve it for her.
But if we hadn’t done that observational research, we wouldn’t have really understood what we were then going to build.
It’s also why we both prefer talking about people rather than personas. A persona easily becomes a category. Real people stay with you.
Cal widens the circle further. In her later years at Fjord, the team started talking about life-centered design, because designing only for humans can harm the environment we depend on. Spending more time in nature, she thinks, keeps that awareness running in the background.
We are animals, we are part of the forest.
Discernment in the Age of AI
Cal’s second piece of advice is about AI itself. It’s a new design tool, and it’s powerful, but its suggestions shouldn’t be accepted at face value. As an advisor to early-stage startups, she sees founders receive forty-page product requirement documents and prototypes stuffed with twenty features they never asked for. The risk is in how our minds respond. We start thinking the extra ideas might be good, and we climb aboard.
You really need to ruthlessly look at everything that AI is suggesting to you and decide whether it makes it into your minimum lovable product or not.
She borrows from Coco Chanel, who advised taking one accessory off before leaving the house, and from Dieter Rams, whose principle was “less but better.”
And the world that we’re in right now with AI, I sort of feel like it’s more but worse.
Part of the problem is that AI has no lived experience to draw on. It only knows the data it was trained on. A founder who goes in with a real, human-grounded idea can easily get pulled in a dozen directions by suggestions that have no roots.
It’s never gone on a research interview into a human home.
Minimum Lovable, Not Just Viable
Many people in startups know they need an MVP, but fewer understand what it should contain. At Fjord, Cal worked with a different frame: the minimum lovable product.
A lovable product still has to work. It helps you finish your task. But it also has some emotional resonance and desirability. Maybe it’s beautifully designed, or something happens that feels a little magical. And it is still edited down to a small number of features, each one clear in its purpose, so people instantly know what the product is for.
Cal is direct about the excuses that creep in. If a confusing kiosk in a parking garage would be called bad design, a confusing app is bad design too, even when the clutter is there for monetization.
Human Metrics
Every designer works inside business goals, and those goals usually mean growth. At Headspace, the tension was sharp. A new product leader once suggested optimizing for time spent in the app. Cal, who ran design, and the colleague who led the science team both pushed back. For a mindfulness product, keeping people on screens was never the point.
Her advice for designers in any business is to advocate for human metrics. If daily, weekly, or monthly active use doesn’t reflect whether your tool is working, find a measure that does.
One thing that I always tell designers is you can advocate for human metrics.
At Headspace, the team tracked stress, anxiety, sleep, and depression as indicators of whether people were really benefiting. She suggested to the team at Nord Security, whose software runs quietly in the background, that they could measure people’s sense of feeling safe. She is especially fond of the SF-36, a 36-question well-being survey from health research with a section on vitality, including whether you feel “full of pep.”
Are we making our users feel more alive?
The psychology and science communities have decades of research on these measures. Cal thinks product teams should borrow from them far more often.
Where User Success Meets Business Success
Designers often feel they can’t move product managers toward human-centered metrics. Cal’s practical answer is to work closely with the data team.
At Headspace, they studied which early behaviors predicted that someone was building a healthy meditation habit. Then they looked at how those behaviors connected to long-term subscriptions. Active days in the first seven and fourteen days turned out to predict retention. The same behavior signaled both a person feeling better and a customer staying.
The lesson came from leaders around her at Headspace, many with human-centered backgrounds from Intuit and IDEO. The user isn’t just at the heart of the product. The user is the customer.
You don’t need to trick the customer, you need to serve the customer.
Even a functional tool can do this, Cal suggests. A ride-hailing app could look at how long tasks take and whether that predicts satisfaction and loyalty.
Skipping to the Last Step
As AI makes building faster, Cal sees teams jump from an idea straight to a high-fidelity prototype. Someone says “wouldn’t it be cool if,” AI builds it, and it goes into an experiment. All the steps that used to edit, ground, and sharpen the idea disappear.
People are skipping all of the steps of the design process except for the very last one, which is like high fidelity building.
The result is a product team, its features, and its customers all spinning together in what I called a washing machine. Nobody knows why something worked or didn’t. Cal also recommends Lou Downe’s work for anyone thinking about this. After Good Services, Downe published Bad Services, along with a video on its fifteen principles.
Painting as a Way Back
So how does a designer stay sober while everyone is a little drunk on AI’s power? For Cal, part of the answer is her art practice.
Painting is physical. She moves around the studio, steps back from the canvas, goes in again. It’s an embodied kind of attention, very different from the forward-facing focus of a screen. Deep focus on a computer can be good too, she says, as long as you monotask. The damage comes from constantly switching between Figma, Slack, and email.
But when I’m painting, there’s no multitasking possible. I’m just painting.
Her painting is also where her personal creative voice comes out. Design is different: it’s a tool in service of others. Keeping those two channels separate makes critique easier, because feedback on a design is never personal.
So if we in a critique are talking through that, it’s not personal.
She thinks this may be why so many designers are signing up for pottery and other hands-on classes. Working with your hands brings you back to earth.
Designing a College
Cal’s latest project is one she clearly loves. She is an advisor to the Greenway Institute, a new engineering college on an old campus in Vermont. Its model aims to let students graduate with little or no debt by spending part of their four years working in paid, entry-level engineering roles. In the US, she notes, the average time to repay student debt is eighteen years.
The business model was already in place when she joined. Her role is service design: mapping the full student journey from arrival to graduation, and finding community partners, like a local gym or performing arts center, so the college can stay lean and focus on teaching. A huge service blueprint now covers a whiteboard in one of their offices. She also used Hyper Island’s Stinky Fish exercise in a workshop there.
Founding president Mark Somerville believes engineering should be learned through hands-on projects, teamwork, and real jobs, with human-centered design built into the curriculum.
It’s really like you must learn by doing.
Students will leave with real work experience, professional skills, and far less debt. For Cal, it’s a reminder of how powerful the design process can be for any problem, not just for apps.
To connect with Cal Thompson, find her on LinkedIn. To learn product design strategy with her, explore Future London Academy.
